In the cloud, data is often shared by a group of users. To ensure the long-term correctness of cloud shared data, a third-party public verifier can be introduced to audit data integrity. During the auditing, protecting the privacy of the contributors of shared data from the public auditor is a fundamental issue. However, this makes it challenging to simultaneously support group membership dynamics efficiently, due to the significant amount of computation needed to update the signatures on shared data. In this paper, we propose a novel privacy-preserving public auditing mechanism for shared cloud data. With our proposed mechanism, a public verifier is able to audit the integrity of shared data without retrieving the entire data from the cloud, and also without learning private identity information of the group members. Group dynamics (user join and user revocation) are efficiently handled by outsourcing signature updating operations to the cloud via a secure proxy re-signature scheme. Experimental results show that our mechanism is highly efficient for dynamic groups.